We revisit the empirical relationship between wages and labor market conditions. Following work histories in the NLSY79 we document that the relationship between wages and unemployment rate differs across occupations. The results hold after controlling for unobserved match quality. This suggests that evidence about history-dependence of wages obtained from pooled samples conceals significant differences and may provide an imprecise description of earning dynamics. Similar discrepancies emerge when we group workers by education. Sensitivity of wages to unemployment appears related to whether total remuneration entails performance pay components.